Intel® oneAPI Math Kernel Library
Ask questions and share information with other developers who use Intel® Math Kernel Library.

Mac High sierra zdotu test not working

jagla__bernd
Beginner
1,405 Views

Hi,

I am struggling with the recompilation of R using the Intel MKL. I have identified the problem to the following:

gfortran -c testf.f 
gcc -c test.c
gcc -L${MKL_LIB_PATH} -o test test.o testf.o -lmkl_intel_ilp64 -lmkl_core -lmkl_sequential  

When running the program, I get a sigfault 11 error message.

I tried debugging with gdb but couldn't really pinpoint the problem. It seems all the good libraries are being used (otool -L) points to the MKL libraries...

I am running out of options...

 

thanks,

 

Bernd

0 Kudos
22 Replies
jagla__bernd
Beginner
178 Views

Thanks for your comments and sorry for the misunderstanding. 

I was able to compile against the Accelerate libraries from Apple and see a speed improvement of about 10x, which is already not that bad.

Do you think I could gain much more with MKL? 

How would you then exchange the libraries? If I am not mistaken, I would have to swap the corresponding libraries in the dynamic libraries that are used. My thought there would try using ar... 

On the other hand, since I already spend a full week on this my time for this project is basically over and I have a working solution. Unless you think that I can gain much more with MKL compared to Accelerate I would rather turn to more important issues on my plate...

Thanks so much for your kind help,

Best,

Bernd

0 Kudos
mecej4
Honored Contributor III
178 Views

A speed increase by a factor of 10 is more than I would have expected, and you should be celebrating it now. Unless you have a cluster with multiple CPUs and/or many threads at your disposal, or you wish to use high performance MKL routines (e.g., Pardiso for solving sparse linear equations) that are not provided by the vendor library that you used, I think that you should put aside the goal of building R with MKL for another time.

You may even find out one day that someone else has built R with MKL, and simply use that version. I think that the maintainers of R should replace the legacy interfacing between C and Fortran by using the Fortran-C interoperability features that are available in Fortran 2008.

0 Kudos
Reply